Efficient intelligent teaching resource scheduling algorithm and implementation method thereof

Abstract

The invention provides a high-efficiency intelligent teaching resource scheduling algorithm and an implementation method thereof, and the method comprises the steps: constructing a judgment matrix according to a learner capability difference evaluation result, cooperation efficiency analysis data, knowledge graph coverage rate statistics and learning style matching degree score, carrying out the calculation to obtain a group heterogeneity index, solving the weight of each index through a feature value method, and carrying out the calculation of the group heterogeneity index. Carrying out weighted summation to obtain a group complementarity comprehensive index; acquiring a historical learning data sequence of a learner, extracting time sequence features, combining a current individual progress deviation monitoring result, individual resource demand evaluation and individual learning content suitability analysis, constructing an individual development trajectory prediction model, and predicting a future individual learning performance curve; and aiming at the group complementarity comprehensive index and a future individual learning performance prediction result, constructing a multi-objective optimization model of resource allocation, setting a group complementarity objective function and an individual satisfaction objective function, and solving to obtain a resource allocation scheme for balancing group and individual benefits.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to an efficient and intelligent teaching resource scheduling algorithm and an implementation method thereof. Background Art

[0002] In the intelligent teaching resource scheduling system, there is a key technical difficulty: how to achieve a dynamic balance between group complementarity and individual development. The system needs to quantify group complementarity through complex algorithms, including calculating the ability difference index between learners, evaluating collaborative effectiveness, analyzing knowledge graph coverage, and learning style matching. At the same time, it is also necessary to identify potential conflicts between individual and group interests, which involves monitoring each learner's progress deviation, resource allocation imbalance, learning content adaptation, and long-term development trajectory. The key challenge is that improving group complementarity may lead to restrictions on the development of some individuals. For example, in order to maximize the overall collaborative effectiveness, the system may assign high-level learners to different groups, but this may slow down the individual progress of these learners. On the other hand, excessive focus on individual needs may reduce the overall benefits of group complementarity. The system needs to constantly adjust and balance between these two conflicting goals. How to evaluate group complementarity indicators and individual development indicators in real time and adjust resource allocation strategies according to changes in these indicators is a technical problem that needs to be solved urgently. Summary of the invention

[0003] The present invention provides an efficient and intelligent teaching resource scheduling algorithm and its implementation method, which mainly includes:

[0004] A judgment matrix was constructed based on the learners' ability difference assessment results, collaborative effectiveness analysis data, knowledge graph coverage statistics, and learning style matching scores. The group heterogeneity index was calculated, and the weights of each indicator were obtained through the eigenvalue method. The weighted sum was then used to obtain the group complementarity comprehensive index.

[0005] Obtain learners' historical learning data sequences, extract time series features, and build an individual development trajectory prediction model based on the current individual progress deviation monitoring results, individual resource demand assessment, and individual learning content adaptability analysis to predict future individual learning performance curves;

[0006] According to the comprehensive index of group complementarity and the prediction results of future individual learning performance, a multi-objective optimization model for resource allocation is constructed, and the group complementarity objective function and the individual satisfaction objective function are set to obtain a resource allocation plan that balances the interests of the group and the individual.

[0007] According to the resource allocation plan, a resource allocation plan evaluation index system is established. The indicators include the improvement of group collaboration efficiency, the release of individual development potential, and the efficiency of group resource utilization. The hierarchical analysis method is used to determine the index weights, calculate the membership matrix of each plan, and obtain the membership calculation results;

[0008] Based on the membership calculation results, determine whether there is a conflict between group complementarity and individual development. If so, continuously optimize the dynamic resource allocation plan, use the group heterogeneity index and individual progress deviation monitoring results as reference factors, and formulate a resource allocation plan that takes into account both group collaboration efficiency and individual resource needs;

[0009] According to the resource allocation plan, a re-evaluation mechanism for the teaching resource node structure is constructed, and resources are subdivided into knowledge point nodes and skill point nodes. The contribution of each node to group complementarity and individual development is evaluated, and the importance of nodes is ranked according to the evaluated contribution. The node resource allocation weights are adjusted, and the knowledge graph coverage and learning style matching indicators are used to dynamically evaluate the balance of node allocation. When the balance is below the threshold, the node resource allocation is optimized to improve the overall efficiency.

[0010] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0011] The present invention discloses an efficient and intelligent teaching resource scheduling algorithm and its implementation method. The method calculates the comprehensive index of group complementarity by constructing a judgment matrix, predicts future performance in combination with individual learning data, and establishes a multi-objective optimization model to balance group and individual interests. For the resource allocation plan, the present invention designs an evaluation index system, and uses the hierarchical analysis method to determine the weight and calculate the membership. When there is a conflict between group complementarity and individual development, the present invention continuously optimizes the resource allocation plan, taking the group heterogeneity index and individual progress deviation as core reference factors. In addition, the present invention also constructs a teaching resource node structure re-evaluation mechanism, sorts the importance and adjusts the weight according to the contribution of the node to group complementarity and individual development, and dynamically evaluates the node allocation balance using indicators such as knowledge graph coverage. The present invention realizes the intelligent allocation of teaching resources and effectively improves the efficiency of group collaboration and individual learning effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 The present invention is a flowchart of an efficient and intelligent teaching resource scheduling algorithm and its implementation method.

[0013] Figure 2 A schematic diagram of an efficient and intelligent teaching resource scheduling algorithm and its implementation method of the present invention.

[0014] Figure 3 This is another schematic diagram of an efficient and intelligent teaching resource scheduling algorithm and its implementation method according to the present invention. DETAILED DESCRIPTION

[0015] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0016] like Figure 1 -3. In this embodiment, an efficient and intelligent teaching resource scheduling algorithm and its implementation method may specifically include:

[0017] Step S101, construct a judgment matrix based on the learners' ability difference assessment results, collaborative effectiveness analysis data, knowledge graph coverage statistics and learning style matching scores, calculate the group heterogeneity index, find the weight of each indicator through the eigenvalue method, and obtain the weighted sum to obtain the group complementarity comprehensive index.

[0018] The individual learning ability distribution data in the learner ability difference assessment results are obtained; according to the individual learning ability distribution data, the learners are divided into three categories: efficient collaboration, medium collaboration and inefficient collaboration; for the three types of learners, the weight coefficients of each evaluation index are determined by using the hierarchical analysis method, and the hierarchical analysis method includes constructing a pairwise comparison matrix, calculating the eigenvector, and performing a consistency test; the group heterogeneity index value is obtained by weighted summation operation; the principal component analysis method is used to reduce the dimension of the group heterogeneity index, and the principal component whose cumulative contribution rate reaches a preset threshold is retained; the radial basis function is selected as the kernel function,

[0019] The penalty parameters are set, and the hyperparameters are optimized by using grid search and cross-validation to establish a population complementarity prediction model; new evaluation data is input into the population complementarity prediction model to calculate the population complementarity comprehensive index value; it is determined whether the population complementarity comprehensive index value exceeds a preset threshold value, and if it exceeds the preset threshold value, the population complementarity level is determined to be high; if it does not exceed the preset threshold value, the population complementarity level is determined to be low; and a population heterogeneity and complementarity analysis report is generated according to the population complementarity level.

[0020] For example, according to the results of learner ability difference assessment, the distribution data of individual learning ability is obtained. The K-means clustering algorithm is used to divide learners into three categories of efficient collaboration, medium collaboration and inefficient collaboration for the collaborative effectiveness analysis data. The Pearson correlation coefficient is calculated for the knowledge graph coverage statistics, and the judgment matrix is ​​constructed in combination with the learning style matching score. The eigenvector is extracted from the judgment matrix and the eigenvalue is calculated. The weight coefficient of each evaluation indicator is determined by the hierarchical analysis method. The specific steps include constructing a pairwise comparison matrix, calculating the eigenvector, and performing a consistency test. The group heterogeneity index value is obtained by weighted summation operation. The principal component analysis method is used to reduce the dimension of the group heterogeneity index, retain the principal component with a cumulative contribution rate of 85%, select the radial basis function as the kernel function, set the penalty parameter C, use grid search and 5-fold cross validation to optimize the hyperparameters, and establish a group complementarity prediction model. New evaluation data is input into the trained group complementarity prediction model to calculate the group complementarity comprehensive index value. The three level thresholds of the group complementarity comprehensive index are set based on historical data, the group complementarity level is judged, and a group heterogeneity and complementarity analysis report is generated. The results of learner ability difference assessment were obtained, including indicators such as cognitive ability, learning speed and knowledge mastery, and the distribution data of individual learning ability was calculated. For the collaborative effectiveness analysis data, the K-means clustering algorithm was used to divide learners into three categories: efficient collaboration, medium collaboration and inefficient collaboration. The initial values ​​of the cluster centers were set to (0.8, 0.5, 0.2), and the calculation was iterated until convergence. The Pearson correlation coefficient was calculated for the knowledge graph coverage statistics to obtain the correlation coefficient matrix R. Combined with the learning style matching score, a 9×9 judgment matrix A was constructed. The eigenvectors were extracted from the judgment matrix A and the maximum eigenvalue λmax=9.45 was calculated. The weight coefficients of each evaluation indicator were determined using the hierarchical analysis method. The specific steps include constructing a pairwise comparison matrix, calculating the eigenvectors, and performing a consistency test (CR=0.037<0.1), and finally obtaining the weight vector W=(0.25, 0.20, 0.15, 0.10, 0.08, 0.07, 0.06, 0.05, 0.04). The population heterogeneity index value GHI=0.72 was obtained by weighted summation. The principal component analysis method was used to reduce the dimension of the population heterogeneity index, and the first three principal components PC1, PC2, and PC3 with a cumulative contribution rate of 85% were retained. The radial basis function (RBF) was selected as the kernel function, the penalty parameter C=1.0 was set, and the hyperparameter γ=0.1 was optimized using grid search and 5-fold cross validation. The support vector machine regression model was established as the population complementarity prediction model. New evaluation data was input into the trained population complementarity prediction model, and the population complementarity comprehensive index value GCI=0.68 was calculated. Based on historical data, the three level thresholds of the population complementarity comprehensive index were set as low 0-0.4, medium 0.4-0.7, and high 0.7-1.0.According to the calculation result GCI=0.68, the group complementarity level is judged to be medium, and an analysis report including heterogeneity index, complementarity index and their interpretation is generated.

[0021] Step S102, obtain the learner's historical learning data sequence, extract the time series features, combine the current individual progress deviation monitoring results, individual resource demand assessment and individual learning content adaptability analysis, build an individual development trajectory prediction model, and predict the future individual learning performance curve.

[0022] The learner's historical learning data sequence is obtained, and the learner's historical learning data sequence is provided by a learning management database; the learner's historical learning data sequence is cleaned and preprocessed to remove abnormal values ​​and missing values; a sliding time window is set, and time series features are extracted according to the sliding time window, and the mean and variance of each learning indicator in the time series features are calculated; the difference between the learner's actual progress and the expected progress is calculated, and the difference is determined by the current individual progress deviation monitoring result; if the difference exceeds a preset threshold, resource demand evaluation and content adaptability analysis are performed; the time series features, progress deviation, resource demand evaluation results and content adaptability analysis results are combined into a feature vector; an individual development trajectory prediction model is constructed using a long short-term memory network, and the long short-term memory network includes an LSTM layer and a fully connected layer of hidden units; feature extraction and nonlinear mapping are performed to obtain a hidden state vector; according to the hidden state vector, a fully connected layer is used to output a future learning performance prediction result; the root mean square error of the learning performance prediction result is calculated; if the root mean square error is lower than a preset accuracy threshold, a stochastic gradient descent algorithm is used to adjust the network parameters; and training is repeated until the root mean square error reaches a preset accuracy threshold.

[0023] Exemplarily, the learner's historical learning data sequence is obtained, the data is cleaned and preprocessed, outliers and missing values ​​are removed, the sliding time window size is set to 30 days, the step length is 1 day, the time series features are extracted, and the mean and variance of indicators such as learning time, number of completed tasks, and test scores in each time window are calculated. Combined with the current individual progress deviation monitoring results, the difference between the learner's actual progress and the expected progress is calculated, and the individual resource demand assessment data is standardized using the Z-score method. The specific steps include calculating the mean and standard deviation of each feature, and then standardizing each data point, calculating the learning content difficulty coefficient according to the pre-defined difficulty level and the learner's historical performance, and performing individual learning content adaptability analysis. The processed time series features, progress deviation, resource demand and content adaptability data are combined into a feature vector as the input of the model. The individual development trajectory prediction model is constructed using a long short-term memory network. An LSTM layer containing 128 hidden units is set, followed by a fully connected layer of 50 neurons. Feature extraction and nonlinear mapping are performed through a multi-layer neural network to obtain a hidden state vector. Based on the hidden state vector, the fully connected layer is used to output the daily learning time, number of completed tasks and test scores in the next 30 days. The prediction accuracy is evaluated by the root mean square error. If the prediction accuracy is lower than the preset threshold of 0.85, the random gradient descent algorithm is used to adjust the network parameters and retrain until the prediction accuracy meets the requirements. Finally, the future individual learning performance curve is generated and the prediction results are presented in the form of a line chart. The learning data of student A in the last 180 days are extracted, including daily learning time, number of completed tasks and test scores. The 3-times standard deviation method is used to eliminate outliers. If the record with a learning time of more than 16 hours is found, it will be deleted. The forward filling method is used to process missing values. The sliding window size is set to 30 days and the step size is 1 day. The features of 150 time windows are calculated. Taking the 100th time window as an example, the average daily learning time is 4.5 hours, with a variance of 0.8 hours; the average number of completed tasks is 6.2, with a variance of 1.5; the average test score is 85 points, with a variance of 25 points. Combined with the individual progress monitoring results, it is found that the current progress of student A is 3 days behind expectations. The resource demand data is Z-score standardized to obtain a standardized resource demand index of 1.2. According to the course difficulty level 1-5 and the historical performance of student A, the difficulty coefficient of the current learning content is calculated to be 3.8. The processed feature vector is input into the LSTM network, which contains an LSTM layer with 128 hidden units and a fully connected layer with 50 neurons. The hidden state vector is obtained by forward propagation. Based on this vector, the fully connected layer is used to predict the learning performance in the next 30 days. After 100 rounds of training, the prediction accuracy reaches 0.88, exceeding the preset threshold of 0.85.The final learning performance curve shows that Student A's average daily study time in the next 30 days is expected to be 5.2 hours, the average number of tasks completed per day is 7.5, and the average test score is expected to reach 88 points, showing a steadily upward trend.

[0024] Step S103, based on the group complementarity comprehensive index and the predicted results of future individual learning performance, a multi-objective optimization model for resource allocation is constructed, a group complementarity objective function and an individual satisfaction objective function are set, and a resource allocation plan that balances the interests of the group and the individual is solved.

[0025] A multi-objective optimization model for resource allocation is constructed based on the comprehensive index of group complementarity and the prediction results of future individual learning performance, wherein a group complementarity objective function f1 and an individual satisfaction objective function f2 are set, wherein f2 is defined as the sum of the squares of the differences between the predicted learning performance and the actual learning performance; the weight coefficients w1 and w2 of the group complementarity objective function and the individual satisfaction objective function are determined by the hierarchical analysis method, and a comprehensive objective function F=w1*f1+w2*f2 is constructed; the population size, crossover probability and mutation probability are set by the genetic algorithm, and single-point crossover and uniform mutation operations are performed to obtain a set of candidate solutions; the candidate solution set is non-dominated sorted, and the Pareto frontier solution is screened out as the final candidate solution set; the candidate solutions are evaluated by the hierarchical analysis method, a judgment matrix is ​​constructed, the eigenvector is calculated, and a consistency test is performed; if the consistency test passes, the final resource allocation plan is determined, and the specific resource allocation ratio of each learner is calculated.

[0026] Exemplarily, according to the comprehensive index of group complementarity and the predicted results of future individual learning performance, a multi-objective optimization model for resource allocation is constructed, and the group complementarity objective function f1 and the individual satisfaction objective function f2 are set, where f2 is defined as the sum of the squares of the difference between the predicted learning performance and the actual learning performance, and the decision variable xi represents the amount of resources allocated to the i-th learner. Constraints such as the total amount of resources not exceeding the preset value and the learning time not exceeding the specified upper limit are established. The weighted summation method is used to transform the multi-objective optimization problem into a single-objective optimization problem. The weight coefficients w1 and w2 of the group complementarity objective function and the individual satisfaction objective function are determined by the hierarchical analysis method, and the comprehensive objective function F = w1f1 + w2f2 is constructed to balance the interests of the group and the individual. The optimization problem is solved by the genetic algorithm, and the population size is set to 100, the crossover probability is 0.8, and the mutation probability is 0.1. The initial population is generated, and the fitness evaluation is performed. The individuals are selected according to the roulette selection method, and the single-point crossover and uniform mutation operations are performed. Iterative optimization obtains a set of candidate solutions. The candidate solution set is non-dominated sorted, and the Pareto frontier solution is screened out as the final candidate solution set. The candidate solutions are evaluated through the hierarchical analysis method, the judgment matrix is ​​constructed, the eigenvector is calculated, the consistency test is performed, the final resource allocation plan is determined, and the specific resource allocation ratio of each learner is calculated.

[0027] Take an online learning platform as an example. The platform has 1000 learners and 100 learning resources. Based on the previously calculated group complementarity comprehensive index of 0.72 and the prediction results of individual learning performance in the next 30 days, a multi-objective optimization model for resource allocation is constructed. The group complementarity objective function f1 is set to maximize the group complementarity index, and the individual satisfaction objective function f2 is defined as the minimization of Σ(predicted learning performance-actual learning performance)^2. The decision variable xi represents the number of resources allocated to the i-th learner. The constraints include that the total number of resources does not exceed 100 and the daily learning time of each learner does not exceed 4 hours. The weight coefficient is determined by the hierarchical analysis method, and w1=0.6 and w2=0.4 are obtained. The genetic algorithm is used to solve the optimization problem, and the population size is set to 100, the crossover probability is 0.8, and the mutation probability is 0.1. After 1000 iterations, 50 non-dominated solutions are obtained. These solutions are Pareto sorted, and 10 Pareto frontier solutions are selected as candidate solutions. A 5×5 judgment matrix was constructed by analytic hierarchy process, and the eigenvector was calculated. The consistency ratio CR=0.03<0.1, and the optimal resource allocation scheme was finally determined. In this scheme, learner A obtained 5 resources, learner B obtained 3 resources, learner C obtained 4 resources, and so on, achieving a balance between group complementarity and individual satisfaction.

[0028] Step S104, based on the resource allocation plan, establish a resource allocation plan evaluation index system, the indicators include the improvement of group collaboration efficiency, the release of individual development potential, and the group resource utilization efficiency, use the hierarchical analysis method to determine the index weights, calculate the membership matrix of each plan, and obtain the membership calculation results.

[0029] A resource allocation scheme evaluation index system is established, and the evaluation index system includes three primary indicators: the degree of improvement of group collaboration efficiency, the degree of release of individual development potential, and the efficiency of group resource utilization; according to the evaluation index system, the index data is standardized by the range normalization method to obtain standardized index data; the weight of each indicator is determined by the hierarchical analysis method, and the characteristic vector is calculated by constructing a judgment matrix to obtain the weight coefficient of the degree of improvement of group collaboration efficiency, the degree of release of individual development potential, and the efficiency of group resource utilization; the fuzzy comprehensive evaluation method is used to construct a single factor evaluation matrix for the standardized index data, and the fuzzy relationship matrix is ​​calculated to obtain the membership of each scheme under different indicators; the membership is converted into a comprehensive evaluation value, and the resource allocation schemes are ranked according to the comprehensive evaluation value to determine the optimal resource allocation scheme.

[0030] For example, according to the resource allocation plan, an evaluation index system is established, including three first-level indicators: the degree of improvement of group collaboration efficiency, the degree of release of individual development potential, and the efficiency of group resource utilization. Ten experts in the field are invited to use a 1-5 point system to quantify and score each indicator. The indicator data is standardized by the extreme difference normalization method. The normalization formula is X' = (X-Xmin) / (Xmax-Xmin). The weight of each indicator is determined by the hierarchical analysis method, and the indicators are compared pairwise using the 1-9 scaling method. A 3x3 judgment matrix is ​​constructed, the eigenvector is calculated, and a consistency test is performed to obtain the weight coefficients of the degree of improvement of group collaboration efficiency, the degree of release of individual development potential, and the efficiency of group resource utilization. The fuzzy comprehensive evaluation method is used to establish an evaluation set. According to historical data, the evaluation results are divided into four levels: excellent, good, general, and poor. A single factor evaluation matrix is ​​constructed, and the fuzzy relationship matrix is ​​calculated to obtain the degree of membership of each plan under different indicators. The membership matrix is ​​converted into a comprehensive evaluation value using the weighted average method. The calculation formula is V = W * R, where W is the weight vector, R is the membership matrix, and V represents the comprehensive evaluation value vector. The various schemes are ranked according to the comprehensive evaluation value, and the optimal resource allocation scheme is obtained. The evaluation report is generated. The report content includes the score of each indicator, the comprehensive evaluation value, and targeted improvement suggestions. Taking the resource allocation scheme evaluation of an online learning platform as an example, an evaluation system including three primary indicators, namely, the improvement of group collaboration efficiency, the release of individual development potential, and the efficiency of group resource utilization, is established. The scores of 10 experts in the field are collected through the questionnaire system, and each indicator is quantified using a 1-5 point system. The collected raw data are normalized to the range. For example, the original score of the improvement of group collaboration efficiency is 4.2, the minimum value is 2.8, the maximum value is 4.5, and the normalized score is 0.82. The hierarchical analysis method is used to determine the indicator weights, and a 3x3 judgment matrix is ​​constructed. The characteristic vector is calculated to obtain the weight coefficients of 0.5 for the improvement of group collaboration efficiency, 0.3 for the release of individual development potential, and 0.2 for the efficiency of group resource utilization. The fuzzy comprehensive evaluation method was used, and the evaluation levels were set as excellent 0.9-1.0, good 0.8-0.9, general 0.7-0.8, and poor 0.6-0.7, and a single factor evaluation matrix was constructed. The membership of a certain scheme under the three indicators was calculated to be 0.85, 0.78, and 0.92, respectively. The comprehensive evaluation value calculated by the weighted average method was 0.844, corresponding to the "good" level. Finally, an evaluation report was generated, including the scores of each indicator, the comprehensive evaluation value, and improvement suggestions for improving the release of individual development potential.

[0031] Step S105, judging whether there is a conflict between group complementarity and individual development based on the membership calculation results, if so, continuously optimizing the dynamic resource allocation plan, taking the group heterogeneity index and individual progress deviation monitoring results as reference factors, and formulating a resource allocation plan that takes into account both group collaboration efficiency and individual resource needs.

[0032] Obtain the membership values ​​of group complementarity and individual development, and determine whether there is a conflict between the membership values ​​according to the preset difference threshold; if the difference between the membership values ​​exceeds the preset threshold, the resource allocation scheme optimization process is triggered. Obtain the monitoring results of the group heterogeneity index and individual progress deviation, and use the group heterogeneity index and individual progress deviation as constraints to construct a multi-objective optimization function. Use a genetic algorithm to solve the multi-objective optimization function, and select non-dominated solutions through non-dominated sorting to obtain a set of candidate resource allocation schemes. Evaluate the set of candidate resource allocation schemes, specifically including establishing an evaluation index system; determining index weights; establishing a fuzzy relationship matrix; and calculating the comprehensive score of each scheme. Select the scheme with the highest score according to the comprehensive score as the optimized dynamic resource allocation scheme. According to the preset optimization cycle, obtain new group heterogeneity index and individual progress deviation data, and re-evaluate and dynamically adjust the dynamic resource allocation scheme.

[0033] Exemplarily, according to the membership calculation results, the membership difference threshold of group complementarity and individual development is set to 0.2 based on historical data statistical analysis. If the difference between the two exceeds the threshold, it is determined that there is a conflict and the resource allocation scheme optimization process is triggered. The monitoring results of the group heterogeneity index and individual progress deviation are obtained, and a multi-objective optimization function f(x) = w1 * group collaboration efficiency + w2 * individual resource demand satisfaction is constructed, where w1 and w2 are weight coefficients, and the group heterogeneity index and individual progress deviation are used as constraints. The genetic algorithm is used to solve the multi-objective optimization problem, and the population size is set to 100, the crossover probability is 0.8, the mutation probability is 0.1, and the maximum number of iterations is 500. The non-dominated solution is screened by non-dominated sorting to form a set of candidate resource allocation schemes. The candidate schemes are evaluated using the fuzzy comprehensive evaluation method, an evaluation index system is constructed, the index weights are determined, a fuzzy relationship matrix is ​​established, the comprehensive scores of each scheme are calculated, and the scheme with the highest score is selected as the optimized dynamic resource allocation scheme. The optimization cycle is set to 7 days, and the resource allocation plan is re-evaluated regularly. Dynamic adjustments are made based on the new group heterogeneity index and individual progress deviation data to continuously optimize the resource allocation strategy. Taking an online learning platform as an example, according to the membership calculation results, it is detected that the group complementarity membership is 0.75, the individual development membership is 0.52, and the difference of 0.23 exceeds the preset threshold of 0.2, triggering the optimization of the resource allocation plan. The current group heterogeneity index of 0.68 and the average individual progress deviation of -3 days are obtained, and the multi-objective optimization function f(x) = 0.6 * group collaboration efficiency + 0.4 * individual resource demand satisfaction is constructed. The genetic algorithm uses this as the fitness function, sets the population size to 100, the crossover probability to 0.8, the mutation probability to 0.1, and the maximum number of iterations to 500. After 300 iterations, it converges and obtains 20 non-dominated solutions. In the fuzzy comprehensive evaluation process, three evaluation indicators were set: group collaboration efficiency, individual resource satisfaction, and learning progress balance, with weights of 0.5, 0.3, and 0.2, respectively. 20 candidate solutions were scored, and the highest scoring solution was 0.82. This solution allocates 60% of resources to group collaboration projects and 40% to individual learning tasks. It was re-evaluated every 7 days. At the end of the first cycle, the group heterogeneity index increased to 0.72, and the average individual progress deviation was reduced to -2 days. Based on this, the resource allocation ratio for the next cycle was dynamically adjusted to 55% group collaboration and 45% individual learning to achieve continuous optimization.

[0034] Step S106: According to the resource allocation plan, a teaching resource node structure re-evaluation mechanism is constructed to subdivide resources into knowledge point nodes and skill point nodes, evaluate the contribution of each node to group complementarity and individual development, rank the importance of the nodes according to the evaluated contribution, and adjust the node resource allocation weight; use the knowledge graph coverage and learning style matching indicators to dynamically evaluate the node allocation balance. When the balance is lower than the threshold, optimize the node resource allocation to improve the overall efficiency.

[0035] The knowledge point and skill point nodes after the teaching resources are segmented are obtained, and a hierarchical structure tree of the nodes is constructed; the nodes in the structure tree are traversed by a depth-first search algorithm, and the contribution of each node is quantified according to the learner's interaction data and learning outcomes; the weight coefficient is determined by the hierarchical analysis method, and a comprehensive score is calculated in combination with the node contribution, and the nodes are ranked by importance; an algorithm is constructed based on the knowledge graph to generate a learning content relationship network, the coverage of the knowledge graph is calculated, and the learning behavior is analyzed to identify the learner's preference; if the coverage of the knowledge graph is lower than a preset threshold, a genetic algorithm is used to optimize the resource allocation ratio of the node; the genetic algorithm includes real number encoding of the node resource allocation ratio, selecting the encoding through a roulette wheel, performing arithmetic crossover and uniform mutation operations on the encoding, and iteratively optimizing the node resource allocation ratio; the optimized node resource allocation ratio is obtained, and the allocation of teaching resources is adjusted.

[0036] Exemplarily, according to the resource allocation plan, the teaching resources are subdivided into knowledge point and skill point nodes, a hierarchical node structure tree is constructed, the depth-first search algorithm is used to traverse the nodes, and the contribution of each node to group complementarity and individual development is quantified based on the learner's interaction data and learning outcomes. The weight coefficients of group complementarity and individual development are determined by the hierarchical analysis method, the comprehensive score is calculated in combination with the node contribution, the importance of the nodes is sorted by the quick sorting algorithm, the sorting position is converted into a weight coefficient through normalization, and the node resource allocation weight is adjusted. Based on the knowledge graph construction algorithm, the learning content relationship network is generated, the knowledge graph coverage is calculated, the learner preference is automatically identified through learning behavior analysis, and the learning style matching degree is calculated using the cosine similarity algorithm. A balance evaluation threshold is set. If the weighted average of the knowledge graph coverage and the learning style matching degree is lower than the threshold, the node resource redistribution process is triggered, and the genetic algorithm is used to optimize the node resource allocation ratio. The encoding method adopts real number encoding, and the fitness function is designed as a balance indicator. Roulette selection, arithmetic crossover and uniform mutation operations are iteratively optimized to improve the overall balance. The evaluation cycle is set to 7 days, and the knowledge graph coverage and learning style matching are recalculated regularly, and the node resource allocation is adjusted dynamically to achieve continuous optimization. Taking an online programming learning platform as an example, the course resources are subdivided into 100 knowledge points and 50 skill points nodes, and a 5-layer deep node structure tree is constructed. The depth-first search algorithm traverses all nodes and calculates the contribution of each node based on the learner's click rate, completion rate and test score in the past 30 days. For example, the group complementarity contribution of the node "loop structure" is 0.85, and the individual development contribution is 0.72. The hierarchical analysis method determines that the group complementarity weight is 0.6 and the individual development weight is 0.4. The calculated comprehensive score of the "loop structure" node is 0.798. The quick sort algorithm sorts 150 nodes, and the "loop structure" ranks 15th. Through normalization, its resource allocation weight is 0.09. The knowledge graph algorithm generates a learning content network containing 1,000 relationships, and the knowledge graph coverage is calculated to be 0.85. By analyzing the learners' stay time and interaction frequency on various resources in the past 7 days, we identified a video preference of 0.7, a text preference of 0.3, and calculated a learning style matching degree of 0.78. The weighted average of the knowledge graph coverage and learning style matching degree of 0.815 was lower than the preset threshold of 0.85, triggering resource redistribution. The genetic algorithm uses equilibrium as the fitness function, a population size of 100, and 500 iterations to finally obtain the optimized node resource allocation plan. The new weight of the "loop structure" node is adjusted to 0.085. Re-evaluate every 7 days and dynamically adjust resource allocation.

[0037] The description of the above embodiments is only used to help understand the technical solutions and core ideas of the present application. Ordinary technical personnel in this field should understand that they can still modify the technical solutions recorded in the aforementioned embodiments, or replace some of the technical features therein with equivalents. However, these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. An efficient and intelligent teaching resource scheduling algorithm and its implementation method, characterized in that: The method comprises: A judgment matrix was constructed based on the learners' ability difference assessment results, collaborative effectiveness analysis data, knowledge graph coverage statistics, and learning style matching scores. The group heterogeneity index was calculated, and the weights of each indicator were obtained through the eigenvalue method. The weighted sum was then used to obtain the group complementarity comprehensive index. Obtain learners' historical learning data sequences, extract time series features, and build an individual development trajectory prediction model based on the current individual progress deviation monitoring results, individual resource demand assessment, and individual learning content adaptability analysis to predict future individual learning performance curves; According to the comprehensive index of group complementarity and the prediction results of future individual learning performance, a multi-objective optimization model for resource allocation is constructed, and the group complementarity objective function and the individual satisfaction objective function are set to obtain a resource allocation plan that balances the interests of the group and the individual. According to the resource allocation plan, a resource allocation plan evaluation index system is established. The indicators include the improvement of group collaboration efficiency, the release of individual development potential, and the efficiency of group resource utilization. The hierarchical analysis method is used to determine the index weights, calculate the membership matrix of each plan, and obtain the membership calculation results; Based on the membership calculation results, determine whether there is a conflict between group complementarity and individual development. If so, continuously optimize the dynamic resource allocation plan, use the group heterogeneity index and individual progress deviation monitoring results as reference factors, and formulate a resource allocation plan that takes into account both group collaboration efficiency and individual resource needs; According to the resource allocation plan, a re-evaluation mechanism for the teaching resource node structure is constructed, and resources are subdivided into knowledge point nodes and skill point nodes. The contribution of each node to group complementarity and individual development is evaluated, and the importance of nodes is ranked according to the evaluated contribution. The node resource allocation weights are adjusted, and the knowledge graph coverage and learning style matching indicators are used to dynamically evaluate the balance of node allocation. When the balance is below the threshold, the node resource allocation is optimized to improve the overall efficiency.

2. The method according to claim 1, wherein: The judgment matrix is ​​constructed based on the learner ability difference assessment results, collaborative effectiveness analysis data, knowledge graph coverage statistics and learning style matching scores, and the group heterogeneity index is calculated. The weight of each indicator is obtained by the eigenvalue method, and the weighted sum is obtained to obtain the group complementarity comprehensive index, including: Obtain individual learning ability distribution data from learners' ability difference assessment results; According to the individual learning ability distribution data, the learners are divided into three categories: efficient collaboration, medium collaboration and inefficient collaboration; For the three types of learners, the weight coefficients of the evaluation indicators are determined by using the hierarchical analysis method, which includes constructing a pairwise comparison matrix, calculating the eigenvector and performing a consistency test; The population heterogeneity index value is obtained through weighted summation operation; The principal component analysis method is used to perform dimensionality reduction processing on the population heterogeneity index, and the principal components whose cumulative contribution rate reaches a preset threshold are retained; The radial basis function was selected as the kernel function, the penalty parameters were set, the hyperparameters were optimized using grid search and cross-validation, and a population complementarity prediction model was established. Inputting new evaluation data into the group complementarity prediction model to calculate the value of the group complementarity comprehensive index; Determining whether the value of the group complementarity comprehensive index exceeds a preset threshold, and if it exceeds the preset threshold, determining that the group complementarity level is high; If the preset threshold is not exceeded, the population complementarity level is determined to be low; A population heterogeneity and complementarity analysis report is generated based on the population complementarity level.

3. The method according to claim 1, wherein: The method of acquiring the learner's historical learning data sequence, extracting the time series features, combining the current individual progress deviation monitoring results, individual resource demand assessment and individual learning content adaptability analysis, constructing an individual development trajectory prediction model, and predicting the future individual learning performance curve includes: Acquire a learner's historical learning data sequence, where the learner's historical learning data sequence is provided by a learning management database; Cleaning and preprocessing the learner's historical learning data sequence to remove abnormal values ​​and missing values; Setting a sliding time window, extracting time series features according to the sliding time window, and calculating the mean and variance of each learning indicator in the time series features; Calculate the difference between the learner's actual progress and the expected progress, the difference determined by the current individual progress deviation monitoring result; If the difference exceeds a preset threshold, resource demand assessment and content suitability analysis are performed; Combining the timing characteristics, progress deviation, resource demand assessment results and content adaptability analysis results into a feature vector; An individual development trajectory prediction model is constructed using a long short-term memory network, wherein the long short-term memory network includes an LSTM layer and a fully connected layer of hidden units; Perform feature extraction and nonlinear mapping to obtain the hidden state vector; Based on the hidden state vector, a fully connected layer is used to output a prediction result of future learning performance; Calculate the root mean square error of the learning performance prediction results; If the RMS error is lower than the preset accuracy threshold, the stochastic gradient descent algorithm is used to adjust the network parameters; The training is repeated until the root mean square error reaches a preset accuracy threshold.

4. The method according to claim 1, wherein: The multi-objective optimization model for resource allocation is constructed based on the group complementarity comprehensive index and the predicted results of future individual learning performance, and the group complementarity objective function and the individual satisfaction objective function are set to obtain a resource allocation plan that balances the interests of the group and the individual, including: A multi-objective optimization model for resource allocation is constructed based on the group complementarity comprehensive index and the prediction results of future individual learning performance, wherein a group complementarity objective function f1 and an individual satisfaction objective function f2 are set, wherein f2 is defined as the sum of the squares of the difference between the predicted learning performance and the actual learning performance; Determine the weight coefficients w1 and w2 of the group complementarity objective function and the individual satisfaction objective function through the hierarchical analysis method, and construct a comprehensive objective function F=w1*f1+w2*f2; The genetic algorithm is used to set the population size, crossover probability and mutation probability, and perform single-point crossover and uniform mutation operations to obtain a set of candidate solutions. Perform non-dominated sorting on the candidate solution set and select the Pareto frontier solution as the final candidate solution set; The candidate solutions were evaluated using the analytic hierarchy process, a judgment matrix was constructed, eigenvectors were calculated, and consistency tests were performed; If the consistency check passes, the final resource allocation plan is determined, and the specific resource allocation ratio for each learner is calculated.

5. The method according to claim 1, wherein: According to the resource allocation plan, a resource allocation plan evaluation index system is established. The indicators include the degree of improvement of group collaboration efficiency, the degree of release of individual development potential, and the efficiency of group resource utilization. The hierarchical analysis method is used to determine the index weights, and the membership matrix of each plan is calculated to obtain the membership calculation results, including: Establishing a resource allocation scheme evaluation index system, which includes three primary indicators: group collaboration efficiency improvement, individual development potential release, and group resource utilization efficiency; According to the evaluation index system, the index data is standardized by using a range normalization method to obtain standardized index data; The weight of each indicator is determined by using the hierarchical analysis method, and the characteristic vector is calculated by constructing a judgment matrix to obtain the weight coefficients of the improvement of group collaboration efficiency, the release of individual development potential, and the efficiency of group resource utilization; Using the fuzzy comprehensive evaluation method, a single factor evaluation matrix is ​​constructed for the standardized indicator data, and the fuzzy relationship matrix is ​​calculated to obtain the degree of membership of each scheme under different indicators; The membership degree is converted into a comprehensive evaluation value, and the resource allocation schemes are ranked according to the comprehensive evaluation value to determine the optimal resource allocation scheme.

6. The method according to claim 1, wherein: The above-mentioned method is to judge whether there is a conflict between group complementarity and individual development based on the membership calculation results. If so, the dynamic resource allocation plan is continuously optimized, and the group heterogeneity index and individual progress deviation monitoring results are used as reference factors to formulate a resource allocation plan that takes into account both group collaboration efficiency and individual resource needs, including: Obtaining membership values ​​of group complementarity and individual development, and judging whether there is a conflict between the membership values ​​according to a preset difference threshold; If the difference between the membership values ​​exceeds a preset threshold, a resource allocation scheme optimization process is triggered; Obtain the monitoring results of group heterogeneity index and individual progress deviation, and use group heterogeneity index and individual progress deviation as constraints to construct a multi-objective optimization function; Genetic algorithm is used to solve multi-objective optimization function, non-dominated solutions are screened through non-dominated sorting, and a set of candidate resource allocation solutions is obtained; Evaluate a set of candidate resource allocation solutions, including establishing an evaluation indicator system; Determine indicator weights; Establish fuzzy relationship matrix; Calculate the comprehensive score of each solution; Selecting the solution with the highest score as the optimized dynamic resource allocation solution according to the comprehensive score; According to the preset optimization cycle, new group heterogeneity index and individual progress deviation data are obtained to re-evaluate and dynamically adjust the dynamic resource allocation plan.

7. The method according to claim 1, wherein: According to the resource allocation plan, a teaching resource node structure re-evaluation mechanism is constructed to subdivide resources into knowledge point and skill point nodes, evaluate the contribution of each node to group complementarity and individual development, rank the importance of nodes according to the evaluated contribution, adjust the node resource allocation weight, and dynamically evaluate the node allocation balance using the knowledge graph coverage and learning style matching indicators. When the balance is lower than the threshold, the node resource allocation is optimized to improve the overall benefit, including: Obtain the knowledge point and skill point nodes after the teaching resources are segmented, and construct a hierarchical structure tree of the nodes; A depth-first search algorithm is used to traverse the nodes in the structure tree, and the contribution of each node is quantified according to the learner's interaction data and learning outcomes; The weight coefficient is determined by using the hierarchical analysis method, and the comprehensive score is calculated based on the node contribution to rank the nodes in importance; Based on the knowledge graph, an algorithm is built to generate a learning content relationship network, calculate the coverage of the knowledge graph, and analyze learning behavior to identify learner preferences; If the knowledge graph coverage is lower than a preset threshold, a genetic algorithm is used to optimize the resource allocation ratio of the nodes; The genetic algorithm includes real number encoding of the node resource allocation ratio, selecting the encoding by roulette, performing arithmetic crossover and uniform mutation operations on the encoding, and iteratively optimizing the node resource allocation ratio; Obtain the optimized node resource allocation ratio and adjust the allocation of teaching resources.

Images